A growing body of research points to a connection between exposure to air pollution and neurodegenerative disorders, including Alzheimer's disease (AD), though mechanisms are not fully understood (Tham and Schikowski, 2021; Parra et al., 2022; Finch, 2023; Franz et al., 2023; Yuan et al., 2023). Various pollutants are present in the air, and some may pose risks to human health. For example, inhalable particulate matter (PM) and nitrogen dioxide (NO₂) have been intensively studied in this regard (Akimoto, 2003; Craig et al., 2008; Dominski et al., 2021). A recent study of UKB (UK Biobank, 2023) data found that higher exposure to PM_2.5 (median particle with diameter ≤ 2.5 μm) and NO₂ was associated with multimorbidity in a dose-dependent manner (Ronaldson et al., 2022). The PM, NO₂, and volatile organic compounds (VOCs) are common components of the traffic-related air pollution (TRAP). These and other types of air pollution (such as ozone, sulfur oxides, carbon monoxide, and lead), might be harmful to the central nervous system (CNS) and promote neuroinflammation and neurodegeneration (Hogan et al., 2015; Calderón-Garcidueñas et al., 2016a; Cheng et al., 2016; Spangenberg and Green, 2017; Costa et al., 2020). A review of epidemiological and experimental studies of the role of PM in neurodegeneration emphasized a link between chronic exposure to PM and onsets of cognitive deficits, dementia, and AD (You et al., 2022). A meta-analysis of 14 studies concluded that PM_2.5 is a risk factor for dementia, with more limited support for nitrogen oxides, though the authors stressed that these results should be interpreted with caution (Wilker et al., 2023). Higher exposure to NO₂ itself was associated with lower cortical thickness of brain regions relevant to AD (Crous-Bou et al., 2020). Another study that used the UKB data (Li et al., 2023) reported an association between residential distance to major roads and dementia that was mediated by TRAP, mainly NO₂.

Exposure to environmental pollutants, including TRAP, could be especially detrimental for hippocampus, a key brain structure for learning and memory, and a primary brain region affected by AD (van der Flier and Scheltens, 2009; Rao et al., 2022). Hippocampal atrophy, manifested in reduced hippocampal volume (HV), is considered one of the major biomarkers of neurodegeneration and preclinical AD pathology (Jack et al., 2018; Grober et al., 2021). It has been associated with a decline in cognitive function and progression of mild cognitive impairment (MCI) to AD (Jack et al., 2000; Henneman et al., 2009; Qu et al., 2023). It was shown that exposure to PM can create profound metabolic disturbances in hippocampus, and adversely affect HV (Park et al., 2020; Balboni et al., 2022). A study that used brain imaging and air pollution data from the UKB found an association between higher PM_2.5 concentration and smaller left HV in adult UKB participants (Hedges et al., 2019).

Genetic factors may also influence HV. For example, carrying the APOE e4 allele (APOE4), the strongest genetic risk factor for AD, may accelerate hippocampal atrophy, along with cognitive decline (Abushakra et al., 2020). Several studies (Tohgi et al., 1997; Reiman et al., 1998; O'Dwyer et al., 2012; Saeed et al., 2021) reported that individuals with APOE4 have markedly smaller HV, along with increased risks of AD and other dementias, compared to those without APOE4. The APOE4 may also interact with exposure to air pollution, including TRAP, potentially modifying its effects on AD-related traits (Schikowski et al., 2015; Ma et al., 2023).

In this study, we used the UKB data to further explore the interactions between APOE4 and TRAP, to better understand how the exposure to TRAP may influence HV in older adults, who carry the strongest genetic risk factor for AD.

We found significant difference in the right HV between groups DNMR and noDNMR, between groups APOE4 and noAPOE4, and between groups DNMR_APOE4 and noDNMR_noAPOE4 for females aged 60–75 years (Table 2, Figure 1). One can see that there was a 0.5% decrease in the right HV for APOE4 carriers, a 1.0% decrease in the right HV for those with DNMR <50, and a 2.5% decrease in right HV for APOE4 carriers with DNMR <50. Note that joint impact of DNMR and APOE4 is larger than separate contributions of APOE4 and DNMR (2.5% > 0.5%, 2.5% > 1.0%), or their sum (2.5% > 0.5% + 1.0%).

There was a 0.6% decrease in the left HV for APOE4 carriers in women aged 60–75 years; differences between other groups were not statistically significant (Supplementary Table 1, Supplementary Figure 1). Normal aging is associated with gradual reducing of HV even without any possible adverse factors (Harman, 2001; Fotuhi et al., 2012; López-Otín et al., 2013). In our analysis, we performed the Welch test to check a possible difference in age between groups (Supplementary Tables 2, 3). We found that on average the subjects in the group DNMR were older than the subjects in the group noDNMR, which might contribute to the reduced HV in the group DNMR compared to the group noDNMR. The subjects in the group APOE4, on average, were younger than the subjects in the group noAPOE4. Note that such age difference between APOE4 and noAPOE4 groups strengthened our results because younger subjects generally tend to have bigger HV than the older ones. The subjects in the group DNMR_APOE4, on average, were the same age as the subjects in the group noDNMR_noAPOE4.

Difference in the right HV between two groups DNMR and noDNMR in the Figure 1A could be attributed to the age only, with participants in the DNMR group older than participants in the noDNMR group (Supplementary Table 3). Based on Table 3, the decrease in the right HV became more pronounced with age in women aged 60–75 years: from 0.8% at age 60 to 0.9% at age 75 for the right HV. Difference in HV between two groups APOE4 and noAPOE4 in the Figure 1B could be attributed to the age and APOE4 carrier status, with participants being younger in the APOE4 compared those in noAPOE4 (Supplementary Table 3). Based on Table 3, the decrease in the right HV became more pronounced with age in women aged 60–75 years: from 0.6% at age 60 to 0.8% at age 75 for the right HV. When taking into account three factors (age, proximity to the nearest major road, and APOE4 carrier status), based on Table 3, difference in the right HV gradually increased from 2.4% at age 60 to 2.8% at age 75, for APOE4 carriers with proximity to the nearest major road <50m. The right HV decreased with age in women aged 60–75 years, losing about 27 mm³/year.

For males aged 60–75 years differences in the left/right HV between studied groups (Supplementary Tables 4, 5) were not statistically significant. For males aged 60–75 years, regression analysis found that for all regression sets Set1 = HV~Age,dnmr (8 models), Set2 = HV~Age,snp (8 models), Set3 = HV~Age,snp,dnmr (64 models) the best (with respect to the minimal AIC criterion) models depend only on Age variable (Supplementary Tables 6–8).

Comparison of the regression model with interactive term with the reference model, i.e., the regression with main additive effects (Table 3, model 13 in Set3: HV~Age,snp,dnmr) allows estimating deviation from the reference model, which was: deviation = 106.2 + 0.4^*Age-(43.3 + 32.7) = 54.2, for Age = 60 and deviation = 60.2, for Age = 75, that is, deviation gradually increases with age. This observation reasonably supported synergy (Roell et al., 2017) in the interaction between DNMR and APOE e4 status with respect to HV decrease.

Our study, using the UKB data, found that female APOE4 carriers aged 60–75 years, who live <50 meters from a major road, had the right HV that was significantly smaller (by about 2.5%) than the HV of the same age women without these conditions. We also showed for the first time that exposure to TRAP (approximated by closeness of participant's main residence to major roads), and carrying the APOE4, synergistically affected HV in women. These findings imply that living farther from major roads may be especially beneficial to older female APOE4 carriers and could help reduce their risks of neurodegenerative disorders, including AD. In our study, the right HV also decreased with age in women aged 60–75 years, losing, on average, about 27 mm³/year. This is in agreement with an earlier report of the HV change with age by the UKB (Nobis et al., 2019).

Results of our study are broadly in line with earlier research that demonstrated that exposure to air pollution (especially to PM_2.5) is associated with smaller brain/hippocampal volume (Wilker et al., 2015; Hedges et al., 2019; Balboni et al., 2022). Several studies investigated the role of air pollution in dementia and cognitive decline, including in APOE4 carriers. Chen et al. (2017) reported a modest increase in hazard ratio (1.07 [95% CI: 1.06–1.08]) of dementia in people living <50 meters from a major road. Higher PM_2.5 exposure was linked to worse cognitive function in APOE4 carriers, but not in non-carriers (Franz et al., 2023). A paper found that associations of PM_2.5, PM₁₀, and NO₂ with cognitive function were more pronounced in female APOE4 carriers (Schikowski et al., 2015). Female APOE4 carriers were also more at risk for air pollution-induced metabolic alterations in hippocampus and cognitive deficits (Calderón-Garcidueñas et al., 2015, 2016b). Another research that used the Women's Health Initiative Memory Study (WHIMS) data found that exposure to a high level of PM_2.5 preceded onset of cognitive impairment in older women, and this relationship varied by APOE genotype, with the largest adverse effect seen in e4/e4 carriers (Cacciottolo et al., 2017). The authors suggested that exposure to PM in the air may accelerate neurodegeneration through various pathways, amyloidogenic, as well as independent of amyloid deposits. A more recent study tested the interaction between APOE genotypes and air pollution and found that the long-term exposure to ambient air pollution was associated with a more rapid cognitive decline in APOE4 carriers (Kulick et al., 2020). Some studies, however, did not find significant interactions between the air pollution and APOE4. For example, a case-control study in northern Taiwan found no differences in susceptibility to air pollution-associated dementia between APOE genotypes (Wu et al., 2015).

One should note that DNMR, which was used as an explanatory variable in our analysis, is an indicator of aggregated exposure to various road-related pollutants, not only to those found in car exhaust fumes. Some of these pollutants may also be potentially relevant to AD pathology. E.g., the higher intensity traffic was associated with the higher concentration of airborne fungi in urban air environments. Examples include Alternaria and Cladosporium species which may cause infection and inflammation, potentially contributing to neurodegeneration (Alonso et al., 2017; Phuna and Madhavan, 2022; Muafa et al., 2024). The role of exposure to airborne fungi in AD pathology deserves separate investigation, especially in the light of our recent findings suggesting that the impact of recurrent fungal infections on AD risk can be larger than that of other types of infections, including bacterial and viral ones (Ukraintseva et al., 2023). Other road-related pollutants, such as noise (The Lancet Regional Health-Europe, 2023), light pollution (Chepesiuk, 2009; Wyse et al., 2011; Aubrecht et al., 2013), and electromagnetic fields (Ahlbom and Feychting, 2003; Kivrak et al., 2017) might also be relevant to health risks. For instance, noise is currently considered a health problem for citizens of the European Union (European Commission, 2023).

We recognize several study limitations. Since only individuals aged 60–75, who have HV measures, were included in the analysis, the sample size in this study was substantially reduced compared to the total UK Biobank sample. Also, different head-size correction (normalization) strategies might yield various volumetric results across studies (Arndt et al., 1991; Mathalon et al., 1993; Goldstein et al., 1999; Seidman et al., 1999; Sanfilipo et al., 2004; Barnes et al., 2010; O'Brien et al., 2011; Voevodskaya et al., 2014). Also, in our study we evaluated regression models using the Akaike information criterion. One should note that there is no universal procedure by which one can determine the “best model”. We applied the AIC approach calculating goodness-of-fit and model variability in order to select the most parsimonious regression model (Burnham and Anderson, 2002; Anderson, 2008; Burnham et al., 2011). Another potential limitation could be that the formal statistical association evaluated from regression analysis may not imply actual causality, which should be further studied using causal inference approaches. Finally, the UK Biobank is volunteer-based study, and so it may not represent general population, therefore, results obtained using this sample should not be extrapolated to the entire UK population, or to other populations, and need further confirmation in additional research.

In summary, this study found that the interaction between APOE4 carrier status and chronic exposure to TRAP (approximated by the closeness of a participant's main residence to a major road) is associated with a significant reduction in hippocampal volume (HV) in female participants of the UK Biobank aged 60–75 years. The results for males didn't reach statistical significance. Our findings suggest that traffic-related air pollution and genetic risk factors for AD (specifically APOE4) can synergistically promote neurodegeneration. Living farther from major roads could help reduce the risks of neurodegenerative disorders, including AD, in older female APOE4 carriers.

This study used de-identified data provided by the UK Biobank (https://www.ukbiobank.ac.uk). This data is not freely available to the public but can be accessed upon approval of a data request by the UK Biobank. Specific policies governing the process to access the UK Biobank data can be found online at https://www.ukbiobank.ac.uk/enable-your-research/apply-for-access.

The studies involving human subjects were approved by the Duke University Health System Institutional Review Board in accordance with the local legislation and institutional requirements. This publication includes only secondary analyses of existing data collected by the UK Biobank and does not include identifiable human data. Written informed consent for the UK Biobank participants was obtained by the UK Biobank (data holder) in accordance with the UK national legislation and the UK Biobank requirements. The latest (at time of calculations) available information on participants' withdrawal in the UK Biobank was taken into account.

VP: Conceptualization, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing – original draft, Writing – review & editing. SU: Conceptualization, Formal analysis, Investigation, Methodology, Project administration, Supervision, Writing – original draft, Writing – review & editing. HD: Data curation, Investigation, Validation, Writing – review & editing. AY: Investigation, Methodology, Writing – review & editing. KA: Investigation, Methodology, Writing – review & editing.